Highlights from the Pierre Auger Observatory · 2012. 12. 20. · Key words: Pierre Auger Observatory, UHECR, cosmic rays 1 Introduction and Status of the Ob-servatory The Pierre

32nd International Cosmic Ray Conference, Beijing 2011

Highlights from the Pierre Auger Observatory

Karl-Heinz Kampert on behalf of the Pierre Auger Collaboration1)

Department of Physics, University of Wuppertal, Germany

Abstract: This paper summarizes some highlights from the Pierre Auger Observatory that were presented

at the ICRC 2011 in Bejing. The cumulative exposure has grown by more than 60% since the previous ICRC

to above 21 000 km2 sr yr. Besides giving important updates on the energy spectrum, mass composition,

arrival directions, and photon and neutrino upper limits, we present first measurements of the energy spectrum

down to 3× 1017 eV, first distributions of the shower maximum, Xmax, together with new surface detectorrelated observables sensitive to Xmax, and we present first measurements of the p-air cross section at ∼ 1018eV. Serendipity observations such as of atmospheric phenomena showing time evolutions of elves extend the

breadth of the astrophysics research program.

Key words: Pierre Auger Observatory, UHECR, cosmic rays

1 Introduction and Status of the Ob-

servatory

The Pierre Auger Observatory started collecting

data in 2004. Since the completion of the base-line

Observatory in 2008 its aperture has grown by about

7000 km2 sr, each year. At this meeting we present

data based on an exposure of more than 21 000 km2

sr yr. The Auger Observatory uses hybrid measure-

ments of air showers recorded by an array of 1660

water Cherenkov surface stations covering an area of

3000 km2, together with 27 air fluorescence telescopes

that observe the development of air showers in the at-

mosphere above the array during dark nights.

An infill array with half the grid size has been

completed and we present first data at this meet-

ing extending the energy spectrum down to 3×1017eV, thereby covering the ankle of the primary energy

spectrum with full detection efficiency. Moreover, the

three high-elevation telescopes (HEAT) started oper-

ation and - together with the infill array in the FOV

of the telescopes - will allow us to extend the hybrid

measurements further down to 1017 eV with unprece-

dented precision. This will enable the study of the

transition from galactic to extra-galactic cosmic rays.

Construction of the buried muon detectors (AMIGA)

in the infill area is in progress. Measurements of the

muons are important for studying the composition

of cosmic rays from surface detector data and their

information is also of vital importance for studying

particle interactions down to the energy of the LHC.

In addition, an extensive R&D program for radio de-

tection of UHE air showers is under way and construc-

tion of the Auger Engineering Radio Array (AERA)

has started. Once completed, it will comprise 160 ra-

dio antennas distributed over an area of 20 km2. First

triple hybrid events composed of particle densities at

ground, longitudinal shower profiles from fluorescence

telescopes, and radio signals from the first anten-

nas have already been observed. Last but not least,

an intense R&D program for microwave detection of

air showers has begun with the first GHz-antennas

presently being installed. These extensions and new

technologies may enhance the performance and capa-

bilities of the Auger Observatory in Argentina and, in

parallel, will explore their potential for a future much

larger ground based observatory.

There are 38 papers presented on behalf of the

Pierre Auger Collaboration at this meeting, and

they are all accessible in five e-print compilations

1)E-mail: auger [email protected]

Vol. 12, 55

K.-H. Kampert: Highlights from the Pierre Auger Observatory

with arXiv numbers 1107.4804, 1107.4805, 1107.4806,

1107.4807, and 1107.4809.

2 The energy spectrum

An accurate measurement of the cosmic ray flux

above 1017 eV is crucial for discriminating between

different models describing the transition between

galactic and extragalactic cosmic rays, the suppres-

sion induced by the cosmic ray propagation, and the

features of the injection spectrum at the sources. Two

complementary techniques are employed at the Pierre

Auger Observatory: a surface detector array (SD)

and a fluorescence detector (FD). The energy spec-

trum at energies greater than 3× 1018 eV has beenderived using data from the SD-array. The analysis

of air showers measured with the FD which also trig-

gered at least one station of the surface detector array

(i.e. hybrid events) enables measurements to be ex-

tended to lower energies. Despite the limited number

of events, due to the fluorescence detector on-time,

the lower energy threshold and the good energy reso-

lution of hybrid events allow us to measure the flux of

cosmic rays with the standard array down to 1018 eV,

into the energy region where the transition between

galactic and extragalactic cosmic rays is expected.

The energy calibration of the SD-array is based

on so-called golden hybrid events, i.e. events that can

be independently reconstructed from the SD and FD

data. Applying high quality cuts, 839 events could be

used for the SD calibration [1, R. Pesce]. The overall

FD energy resolution is 7.6% and it is almost con-

stant with energy. The total systematic uncertainty

on the FD energy scale is about 22%. It includes con-

tributions from the absolute fluorescence yield (14%),

calibration of the fluorescence telescopes (9.5%), the

invisible energy correction (4%), systematics in the

reconstruction method used to calculate the shower

longitudinal profile (10%), and atmospheric effects

(6% - 8%). The atmospheric uncertainties include

those related to the measurements of aerosol optical

depth (5% - 7.5%), phase function (1%) and wave-

length dependence (0.5%), the atmosphere variability

(1%) and the residual uncertainties on the estimation

of pressure, temperature and humidity dependence of

the fluorescence yield (1.5%).

The energy spectrum derived from hybrid data

has been combined with the one obtained from

surface detector data using a maximum likelihood

method and is shown in Figure 1 together with a bro-

ken power law and a smooth cut-off at higher energies

[1, F. Salamida]. Both, the ankle and suppression of

the flux at higher energies are clearly visible. The

spectrum can be compared to astrophysical models

and can be described by both a proton and heavy-

dominated composition at the highest energies. Thus,

measurements of the composition are needed to dis-

criminate between various astrophysical models.

Fig. 1. Combined energy spectrum fitted with

two power laws in the ankle region and a

smoothly changing function at higher energies.

Only statistical uncertainties are shown. The

systematic uncertainty in the energy scale is

22%.

Data of the 750 m infill array reach full efficiency

for all primaries at E > 3×1017 eV and, using datawith an exposure of 26 km2 sr yr, extend the spec-

trum of Figure 1 smoothly down to this threshold [1,

I. Maris]. Analysis of the composition in this energy

range is on-going. HEAT data, combined with the

infill array, extend the energy range further down to

1017 eV [2, H. J. Mathes].

3 The cosmic ray mass composition

As discussed above, measuring mass composition

of cosmic rays along with the flux is a key to sep-

arating the different scenarios of origin and propa-

gation of cosmic rays. The composition must be in-

ferred from measurements of various shower observ-

ables, most importantly the atmospheric depth at

which the shower attains its maximum size, Xmax.

For a given shower, the position of Xmax will depend

on the depth of the first interaction of the primary in

the atmosphere and the depth that it takes the cas-

cade to develop. Thereby, it will depend not only on

the primary mass, but also on the cross section of the

primary particle with air and on features of hadronic

interactions at high energies. This important caveat

should be kept in mind when discussing the mass

Vol. 12, 56


composition of cosmic rays, i.e. interpretation of

shower observables in terms of primary mass are sub-

ject to deficiencies of hadronic interaction models em-

ployed in air shower simulations. Besides the position

of Xmax, its fluctuations on shower-by-shower basis,

RMS(Xmax), show strong sensitivity to the primary

mass.

For the analysis, again hybrid data are used and

the shower profiles are required to be good fits to a

Gaisser-Hillas function, as deviations could indicate

the presence of residual clouds. Both 〈Xmax〉 and itsRMS show a characteristic change at E>∼ 5×1018 eVindicating an increasingly heavier composition when

compared to EAS simulations [3, P. Facal] (c.f. Figure

3). It is well known that MC predictions are more un-

certain for the 〈Xmax〉 than for the fluctuations. Thisis mainly due to the additional dependence of 〈Xmax〉on the multiplicity in hadronic interactions. In Fig-

ure 2 we therefore compare the shape of the distribu-

tions, Xmax−〈Xmax〉, to MC predictions for differentcompositions and hadronic interaction models. As

can be seen, in this representation the various models

Fig. 2. Centered distribution, Xmax −〈Xmax〉,for the lowest and highest energy bins. Sub-

traction of the mean allows the comparison of

the shapes of these distributions with the su-

perimposed MC simulations.

predict a nearly universal shape. At low energy, the

shape of the distribution is compatible with a very

light or mixed composition, whereas at high energies,

the narrow shape would favor a significant fraction of

nuclei (CNO or heavier).

Fluorescence telescopes are the only observational

tool currently enabling direct measurements of the

shower maximum Xmax. Unfortunately, those data

suffer from statistics because of the duty cycle being

only ∼ 15%. However, surface detectors, operated24 hours a day, also provide observables which are

related to the longitudinal shower profile. These ob-

servables are subject to independent systematic un-

certainties (both experimentally and theoretically).

The higher statistics allow us to extend these mea-

surements to higher energies than possible with the

FD.

For each SD event, the water-Cherenkov detec-

tors record their signals as a function of time. Since

muons travel in almost straight lines whereas the elec-

tromagnetic particles suffer more multiple scattering

on their way to ground, the first part of the signal is

dominated by the muon component. Due to the ab-

sorption of the electromagnetic (EM) component, the

number of these particles at the ground depends, for

a given energy, on the distance to the shower maxi-

mum and therefore on the primary mass. In conse-

quence, the time profile of particles reaching ground is

sensitive to the cascade development: the higher the

production height, the narrower the time pulse. The

time distribution of the SD signal is characterised by

means of the risetime, t1/2, which depends on the dis-

tance to the shower maximum, the zenith angle θ and

the distance to the core r. In a first step, the zenith

angle at which the risetime asymmetries between the

inner and outer SD stations of a shower become max-

imal is is analyzed. This angle Θmax is then, in a

second step, related to the shower maximum using a

subset of hybrid events [3, D. Garcia-Pinto]. Using

this correlation it is possible to measure the shower

evolution with surface detector data, in a similar way

as to that with the SD energy calibration is performed

for a subset of events with the FD data. The result

is shown in Figure 3.

Not only the risetime of signals in the SD tanks,

but also the arrival time of particles with respect to

the shower front plane contains information about

the position of the shower maximum. A method for

reconstructing the socalled Muon Production Depth

(MPD), i.e. the depth at which a given muon is pro-

duced, measured parallel to the shower axis, using the

FADC traces of detectors far from the core, has been

Vol. 12, 57


Fig. 3. Results on shower evolution sensitive

observables compared with models prediction.

The error bars correspond to the statistical

uncertainty. The systematic uncertainty is

represented by the shaded bands.

presented in [4]. From the MPDs an observable can

be defined, Xμmax, as the depth along the shower axis

where the number of produced muons in a shower

reaches a maximum. The method is currently re-

stricted to inclined showers where muons dominate

the signal at ground level. Once the MPDs are ob-

tained for each event, the value of Xμmax is found by

fitting a Gaisser-Hillas function to the depth profile.

The results of 〈Xμmax〉 presented in Figure 3 are re-stricted to zenith angles between 55◦ and 65◦ and use

timing information only for detectors far from the

core (r > 1800 m). Because of this distance restric-

tion, the effective energy range for which the method

can presently be applied is limited to E> 2×1019 eV.The measured values of 〈Xμmax〉 are presented in theupper panel of Figure 3. It is important to point out,

that the predictions of Xμmax from different hadronic

models would not be affected if a discrepancy between

a model and data [3, J. Allen] is limited to the total

number of muons. However, differences in the muon

energy and spatial distribution would modify the pre-

dictions.

With this caveat concerning hadronic interaction

models, one might infer the primary composition from

the data on the longitudinal air shower development

presented in Figure 3. The evolution of 〈Xmax〉, Θmaxand 〈Xμmax〉 with energy is similar, despite the factthat the three analyses come from completely inde-

pendent techniques that have different sources of sys-

tematic uncertainties. Concerning the RMS of Xmax,

a variety of compositions can give rise to large values

of the RMS, because the width of the Xmax is in-

fluenced by both, the shower-to-shower fluctuations

of individual components and their relative displace-

ment in terms of 〈Xmax〉 [5]. However, within exper-imental uncertainties, the behaviour of 〈Xmax〉, Θmaxand 〈Xμmax〉 as shown in Figure 3 is compatible withthe energy evolution of RMS(Xmax). In particular,

at the highest energies all four analyses show consis-

tently that our data better resemble the simulations

of heavier primaries than pure protons.

4 p-air cross section and tests of

hadronic interaction models

One of the biggest challenges for a better under-

standing of the nature of ultra-high energy cosmic

rays is to improve the modeling of hadronic inter-

actions in air showers. None of the current mod-

els is able to consistently describe cosmic ray data,

which most importantly prevents a precise determi-

nation of the primary cosmic ray mass composition.

Studies to exploit the sensitivity of cosmic ray data

to the characteristics of hadronic interactions at en-

ergies beyond state-of-the-art accelerator technology

began over 50 years ago. The property of interactions

most directly linked to the development of extensive

air showers is the cross section for the production

of hadronic particles (e.g. [6]). To reconstruct the

proton-air cross-section based on hybrid data, we an-

alyze the shape of the distribution of the largest val-

ues of the depth of shower maxima, Xmax. This tail of

the Xmax-distribution, that contains the 20% of deep-

est showers, exhibits the expected exponential shape

Vol. 12, 58


dN/dXmax ∝ exp(−Xmax/Λ20). It is directly relatedto the p-air cross section via σp−air = 〈mair〉/Λ20. Inpractice, to properly account for shower fluctuations

and detector effects, the exponential tail is compared

to Monte Carlo predictions. Any disagreement be-

tween data and predictions is then attributed to a

modified value of the proton-air cross-section [3, R.

Ulrich]. In this analysis, the energy interval is re-

stricted to 1018 to 1018.5 eV which corresponds to a

center-of-mass energy in the nucleon-nucleon system

of√s = 57 TeV. This interval has been chosen be-

cause of high statistics in the data and because of the

composition being compatible with a dominance of

protons (see. Sec. 3). A possible contamination of

He primaries could mimic a larger cross section (e.g.

by 20 mb for 20% He contamination) while a photon

contamination could reduce the cross section by at

most 10 mb. Combining the results one finds

σp−air =(505±22stat(+19−14)syst

)mb

at a center-of-mass energy of 57±6 TeV. This resultis shown in comparison to other data and models in

Figure 4.

Fig. 4. Proton-air cross section compared to

other measurements and model predictions

(for references see [3, R. Ulrich]). The inner

error bars are statistical only, while the outer

include all systematic uncertainties for a he-

lium fraction of 25% and 10 mb photon sys-

tematics.

The result favor a moderately slow rise of the cross

section towards higher energies, well in line with re-

cent results from LHC (e.g. [7]). A conversion of the

derived σp−air measurement into the more fundamen-

tal cross-section of proton-proton collisions using the

Glauber framework [8, 9] will be published elsewhere.

The importance of hadronic interaction models

to measurements of the cosmic ray mass composi-

tion has been addressed in Secttion 3. In particular,

muons in extensive air showers are subject to large

theoretical uncertainties due to our limited knowledge

of multi-particle production in hadronic interactions.

However, hybrid data can be used to constrain the

models and to uncover deficiencies in describing fea-

tures of EAS data. When measuring the longitudi-

nal profile (LP) in a golden hybrid event, we con-

struct a library of simulated air-shower events with

the same shower geometry where the LP of each sim-

ulated event matches a measured one. The measured

LP constrains the natural shower-to-shower fluctua-

tions of the distribution of particles at ground. This

allows the ground signals of simulated events to be

compared to the ground signals of measured events

on an event-by-event basis. An example of such an

analysis is shown in Figure 5 [3, J. Allen]. Here, the

LP of a measured event is compared to p and Fe

simulations, each providing a good fit to the data.

The bottom panel shows the corresponding signals

in the SD. The ratio of the measured signal at 1000

m from the shower core, S(1000), to that predicted

in simulations of showers with proton primaries,

Fig. 5. Top: A longitudinal profile measured

for a hybrid event and matching simulations of

two showers with proton and iron primaries.

Bottom: A lateral distribution function deter-

mined for the same hybrid event as in the top

panel and that of the two simulated events.

Vol. 12, 59


S(1000)DataS(1000)Sim

, is 1.5 for vertical showers and grows to

around 2 for inclined events. The ground signal of

more-inclined events is muon-dominated. Therefore,

the increase of the discrepancy with zenith angle sug-

gests that there is a deficit of muons in the simulated

showers compared to the data. The discrepancy ex-

ists for simulations of showers with iron primaries as

well, which means that the ground signal cannot be

explained only through composition.

This finding is corroborated by direct estimations

of the muon number from the signal traces in the

SD as well as by making use of universality of the

muonic to electromagnetic signal Sμ(1000)/Sem(1000)

for fixed vertical depth of the shower [3, J. Allen].

Moreover, this purely observational estimation of the

muonic signal in data, is compatible with results ob-

tained from inclined showers [3, G. Rodriguez]. This

is best illustrated by Figure 6. Here, N19 is defined

as the ratio of the total number of muons, Nμ, in the

shower with respect to the total number of muons

at E = 10 EeV given by a 2-dim reference distribu-

tion, N19 = Nμ(E,θ)/Nmapμ (E = 10EeV,θ) which ac-

counts for the geomagnetic spatial deviation of muons

at ground. Thus, all of the analyses show a significant

deficit in the number of muons predicted by simula-

tions with proton primaries compared to data. This

discrepancy cannot be explained by the composition

alone, although a heavy composition could reduce the

relative excess by up to 40%. The increased sophis-

tication of the methods gives further weight to the

Fig. 6. Fit of a calibration curve N19 =

A(E/10EeV)B . The constants A and B

are obtained using the maximum likelihood

method. The contours indicate the constant

levels of the p.d.f. integrated over zenith an-

gle, corresponding to 10, 50 and 90% of the

maximum value [1, H. Dembinski]. Calibra-

tion curves for protons QGSJETII (dot line)

and iron EPOS1.99 (dashed line) are shown

for comparison.

conclusions that, at the current fluorescence en-

ergy scale, the number of muons in data is nearly

twice that predicted by simulations of proton-induced

showers. The possible zenith angle dependence of

N relμ suggests that, in addition to the number, there

may also be a discrepancy in the attenuation and lat-

eral distribution of muons between the simulations

and data.

5 Update of photon and neutrino up-

per limits

The search, and possibly study, of high energy

photons and neutrinos is of interest for at least three

reasons: (i) Top- Down models of UHECR origin [10]

including topological defect or super-heavy dark mat-

ter models predict a significant fraction of photons

and neutrinos at the highest energies, (ii) they would

provide a smoking-gun signature of the GZK-effect

because of the decay of charged and neutral pions cre-

ated in photo-pion production, and (iii) they would

open a new window to the most extreme Universe by

possibly seeing point sources in the sky. A search

for their signatures has thus been part of the Auger

research program from the beginning.

The search for EeV photons presented at this

meeting is based on hybrid events. Due to the FD

duty cycle the event statistics is reduced compared

to the SD-only detection mode. However, the hybrid

detection technique provides a precise geometry and

energy determination with the additional benefit of

allowing to reduce the energy threshold for detection

to about 1018 eV. To improve the photon-hadron dis-

crimination power over measurements of Xmax only,

the differences in the lateral distribution functions for

photons and hadrons measured by the SD have been

considered by analyzing the observable, Sb, defined

in [11]. To reject misreconstructed profiles, only time

periods with the sky not obscured by clouds, and with

a reliable measurement of the vertical optical depth

of aerosols, are selected. On the SD side we require

at least 4 active stations within 2 km of the hybrid

reconstructed axis. This prevents an underestimation

of Sb (which would mimic the behavior of a photon

event) due to missing or temporarily inefficient de-

tectors. For the classification of photon candidates,

a Fisher analysis trained with a sample of a total of

∼ 30000 photon and proton CORSIKA [12] showersgenerated according to a power law spectrum between

1017 and 1020 eV is performed [13, M. Settimo]. The

Fisher response distributions for photon and proton

Vol. 12, 60


primaries are well separated for all energies above 1018

eV. Photon-like events in the data are then selected

by applying an “a priori” cut to the upper 50% of the

photon like events. This reduces the photon detection

efficiency to 50% but provides a conservative result

in the upper limit calculation by reducing the depen-

dence on the hadronic interaction models and on the

mass composition assumption. With this choice, the

expected hadron contamination is about 1% in the

lowest energy interval (between 1018 and 1018.5 eV)

and it becomes smaller for increasing energies [13, M.

Settimo].

Applying the method to data, 6, 0, 0, 0 and 0

photon candidates are found for energies above 1, 2,

3, 5 and 10 EeV. We checked with simulations that

the observed number of photon candidates is consis-

tent with the expectation for nuclear primaries, under

the assumption of a mixed composition. The corre-

sponding 95% CL upper limits on the photon flux

Φ95CLγ integrated above an energy threshold E0 are

shown in Figure 7. To be conservative, a minimum

value of the exposure above E0 is used and a possible

nuclear background is not subtracted for the calcula-

tion of N95CLγ . The flux limits shown in Figure 7 or

likewise the derived limits on the photon fraction of

0.4%, 0.5%, 1.0%, 2.6% and 8.9% above 1, 2, 3, 5 and

10 EeV, significantly improve previous results at the

lower energies and rule out exotic models of UHECR

origin, except for the Z−burst model of Ref. [14].While the focus of the current analysis was the low

EeV range, future work will be performed to improve

the photon-hadron separation also at higher energies

Fig. 7. Upper limits on the photon flux above

1, 2, 3, 5 and 10 EeV derived in this work

(red arrows) compared to previous limits from

Auger, from AGASA (A), and Yakutsk (Y).

The shaded region and the lines give the pre-

dictions for the GZK photon flux and for

top-down models (TD, Z-Burst, SHDM and

SHDM’). (See [13, M. Settimo] for references.)

using further information provided by the SD.

The surface detector is well suited also to search

for ultrahigh energy neutrinos in the sub-EeV energy

range and above. Neutrinos of all flavours can in-

teract in the atmosphere and induce inclined show-

ers close to the ground (down-going). The sen-

sitivity of the SD to tau neutrinos is further en-

hanced through the “Earth-skimming” mechanism

(up-going). Both types of neutrino interactions can

be identified through the broad time structure of the

signals induced in the SD stations.

The analysis starts with the inclined shower se-

lection (down-going: θ > 75◦ and Earth-skimming

θ < 96◦). These showers usually have elongated pat-

terns on the ground along the azimuthal arrival di-

rection. A length L and a width W are assigned

to the pattern and a cut on their ratio L/W is ap-

plied. We also calculate the apparent speed V of an

event using the times of signals at ground and the

distances between stations projected onto L. Finally,

for down-going events, we reconstruct the zenith an-

gle θrec. After this pre-selection, the FADC signal-

traces of the SD stations are analyzed to search for

so-called “young showers” with a broad time struc-

ture. To optimize the discrimination power, again a

Fisher discriminant method is used and trained to a

subset of data. The identification efficiency for the set

of selection cuts applied to the data depends on the

neutrino energy Eν , the slant depth D from ground to

the neutrino interaction point, the shower geometry,

the neutrino flavour (νe, νμ, or ντ ), and is different

for CC- and NC-type interactions, see [15].

Using the independent sets of identification crite-

ria that were designed to search for down- and up-

going neutrinos in the data collected from 1 Jan-

uary 2004 to 31 May 2010, no candidate was found

[13, Y. Guardincerri]. Assuming a differential flux

f(Eν) = kE−2, we place a 90% CL upper limit on

the single flavour neutrino flux of k < 3.2×10−8 GeVcm−2 s−1 sr−1 in the energy interval 1.6× 1017 eV−2.0× 1019 eV, based on Earth-skimming neutrinosand k < 1.7×10−7 GeV cm−2 s−1 sr−1 in the energyinterval 1×1017 eV −1×1020 eV, based on down-goingneutrinos (see Figure 8). The optimistic fluxes for p-

primaries shown in this figure are accessible for the

proposed lifetime of the Pierre Auger Observatory.

The transition models and sources with a dominance

of heavy primaries would be challenging to reach for

any of the currently operating experiments.

With no candidate events found in the search pe-

riod, we can also place a limit on the UHE neutrino

flux from a source at declination δ. Since the sensi-

Vol. 12, 61


Fig. 8. Differential upper limits (90% CL per

half decade of energy) from the Pierre Auger

Observatory for a diffuse flux of down-going

ν (2 yr of full Auger) and Earth-skimming ντ(3.5 yr of full Auger [13, Y. Guardincerri]).

Limits from other experiments and expected

fluxes are also shown (see [5] for references).

tivity to UHEνs is limited to large zenith angles, the

rate of events from a point source in the sky depends

strongly on its declination. In both Earthskimming

and down-going analyses the sensitivity yields a broad

“plateau” spanning Δδ∼ 110◦ in declination with thehighest sensitivity reached at δ ∼+55◦. The presentflux limits do not yet allow us to constrain models of

UHEν production in the jets and the core of CenA

[16].

6 Anisotropies

One of the keys to understanding the nature of

UHECRs is their distribution over the sky. This

distribution depends on the location of the UHECR

sources, as well on the UHECR mass composition and

large-scale magnetic fields, both Galactic and extra-

galactic. Despite significant efforts, none of these is-

sues is well understood at present. Observation of

the suppression of the CR flux at the highest ener-

gies (c.f. Secttion 2) and its interpretation in terms

of the GZK effect suggests that the closest sources

of UHECRs are situated within the GZK volume of

dGZK 5.5×1019 eV with AGN from the Veron-Cetty-Veron cat-

alog [18] within 75 Mpc on an angular scale of 3.1◦

at the 99% CL. The optimal parameters were found

using a exploratory scan (Period I) and independent

data (Period II) showed 8 of 13 events correlating. An

update then yielded 21 of 55 events (Period II+III)

correlating for the same parameter set. Here, we

present the latest update including data up to June

2011 (c.f. Figure 9) which yields a total of 28 of 84

events (Period II+III+IV) showing a correlation on

a 3.1◦-scale with a nearby AGN. The overall correla-

tion strength thus decreased from (62±10) initially to(33±5)%. The chance probability of observing sucha correlation from a random distribution remains be-

low 1%. Cumulative plots are often misleading and

Figure 9 may be interpreted as a signal that is fading

away. Thus, the superimposed black symbols show in

addition the averages of 10 independent consecutive

events. Obviously, the first bin is an upwards fluctu-

ation by about 3σ from the mean of all events while

the rest of the dataset does not show any peculiarity.

Evidently, more data is needed to arrive at a definite

conclusion.

Fig. 9. The most likely value of the degree of

correlation pdata = k/N is plotted as a func-

tion of the total number of time-ordered events

(excluding those in period I). The 68%, 95%

and 99.7% confidence level intervals around

the most likely value are shaded. The hori-

zontal dashed line shows the isotropic value

piso = 0.21 and the full line the current esti-

mate of the signal pdata = 0.33± 0.05. Theblack symbols show the correlation fractions

bins of independent 10 consecutive events.

Interestingly, at this meeting the Telescope Ar-

ray Collaboration presented, though with a much

lower exposure of ∼ 20% of Auger, an analysis of

Vol. 12, 62


the northern sky adapting the parameters from the

Auger collaboration (for a recent update see [19]).

With 11 correlating events of 25 being above their

energy threshold, they find a signal strength of 44%.

Correcting this for the larger chance probability of

piso = 0.24 compared to 0.21 in Auger, a good agree-

ment of the data sets can be concluded. However,

the TA events alone can originate from an isotropic

distribution with a chance probability of about 2%.

The sky region around Centaurus A is populated

by a larger number of high energy events compared

to the rest of the sky, with the largest departure from

isotropy at 24◦ around the center of Cen A with 19

events observed and 7.6 expected for isotropy. How-

ever, a Kolomogorov-Smirnov test shows a chance

probability for this to occur at a level of 4%. Sim-

ilarly, a search for directionally-aligned events (or

“multiplets”) expected from sets of events coming

from the same source after having been deflected by

intervening coherent magnetic fields shows one 12-

plet with an energy threshold of 20 EeV. The prob-

ability that it appears by chance from an isotropic

distribution of events is again 6%. Thus, there is

no significant evidence for the existence of correlated

multiplets in the present data set [13, G. Golup] and

[20]. It will be interesting to check if some of the

observed multiplets grow significantly or if some new

large multiplets appear. If one of them were a real

multiplet, doubling the statistics should double its

multiplicity, i.e. the significance does not increase as√N but much faster.

The Pierre Auger Observatory has sensitivity also

to neutron fluxes produced at cosmic ray acceleration

sites in the Galaxy. Because of relativistic time di-

lation, the neutron mean decay length is (9.2×E)kpc, where E is the neutron energy in EeV. A blind

search over the field of view of the Auger Observatory

for a point-like excess yields no statistically significant

candidates. The galactic center is a particularly in-

teresting target because of the presence of a massive

black hole. The results for the window centered on it

and for E ≥ 1 EeV shows no excess with a 95% CLupper limit on the flux from a point source in this

direction of 0.01 km−2 yr−1 [13, B. Rouille d’Orfeuil],

which updates the bounds obtained previously [21].

We note that for directions along the Galactic plane

the upper limits are below 0.024 km−2 yr−1, 0.014

km−2 yr−1 and 0.026 km−2 yr−1 for the energy bins

[1−2] EeV, [2−3] EeV and E≥ 1 EeV, respectively.A targeted search has also been performed to

test potential sources of galactic cosmic rays, such as

SNR, pulsars and Pulsar Wind Nebula (PWN). The

candidate sources are expected to be strong gamma-

ray emitters at GeV and TeV energies. For this

reason, we apply a neutron search also to Galactic

gamma-ray sources extracted from the Fermi LAT

Point Source Catalog [22] and the H.E.S.S. Source

Catalog∗. Targets were selected among the sources

located in the portion of the Galactic plane, defined

as |b| < 10◦, covered by the FOV of the SD, and lo-cated at a distance shorter than 9 kpc (λn at 1 EeV)

[13, B. Rouille d’Orfeuil]. The stacked signal com-

puted from the SD data at the positions of the two

sets of sources under study and for the same energy

bins used in the Galactic plane search has not yet

yielded an excess.

Besides searching for point sources of charged cos-

mic rays or neutrons, the large scale distribution of

arrival directions of CRs represents another impor-

tant tool for understanding their origin. Using data

from the SD array, upper limits below 2% at 99%

CL have been recently reported for EeV energies on

the dipole component in the equatorial plane [23].

Such upper limits are sensible, because cosmic rays

of galactic origin, while escaping from the galaxy in

this energy range, might generate a dipolar large-scale

anisotropy with an amplitude at the % level as seen

from the Earth. Even for isotropic extragalactic cos-

mic rays, a large scale anisotropy may remain due to

the motion of our galaxy with respect to the frame

of extragalactic isotropy. This anisotropy would be

dipolar in a similar way to the Compton-Getting ef-

fect [24] in the absence of the galactic magnetic field.

An update of the results of searches for first har-

monic modulations in the right ascension distribution

of cosmic rays is presented in Figure 10 [13, H. Ly-

beris]. The upper limits at 99% CL obtained here

provide the most stringent bounds at present above

2.5×1017 eV. Some predictions for anisotropies aris-ing from models of both galactic and extragalactic

cosmic ray origin are included in the plot together

with data from other experiments. In models A and

S (A and S standing for 2 different galactic mag-

netic field symmetries [25]), the anisotropy is caused

by drift motions due to the regular component of the

galactic magnetic field, while in model Gal [26], the

anisotropy is caused by purely diffusive motions due

to the turbulent component of the field. Some of these

amplitudes are challenged by our current sensitiv-

ity. For extragalactic cosmic rays considered in model

C−G Xgal [27], the motion of our galaxy with respectto the CMB (supposed to be the frame of extragalac-

∗http://www.mpi-hd.mpg.de/hfm/HESS/pages/home/sources/

Vol. 12, 63


tic isotropy) induces the small dipolar anisotropy (ne-

glecting the effect of the galactic magnetic field).

While the measurements of the amplitudes do not

provide any evidence for anisotropy, it is interest-

ing to note that the phase shown in Figure 11 sug-

gests a smooth transition between a common phase of

� 270◦ below 1 EeV and another phase (right ascen-sion � 100◦) above 5 EeV. This is potentially inter-esting, because, with a real underlying anisotropy, a

consistency of the phase measurements in ordered en-

ergy intervals is indeed expected with lower statistics

than that required for the amplitudes to significantly

stand out of the background noise. Applying a Likeli-

hood test leads to a probability of ∼ 10−3 of observingthis from a random distribution. However, since we

Fig. 10. Upper limits on the anisotropy : equa-

torial dipole component d⊥ as a function of

energy from Auger. Results from EAS-TOP,

AGASA, KASCADE and KASCADE-Grande

experiments are also displayed, in addition to

several predictions (see [13, H. Lyberis] for ref-

erences.

Fig. 11. Phase of the first harmonic as a func-

tion of energy. The dashed line, resulting from

an empirical fit, is used in the likelihood ratio

test (see text) [13, H. Lyberis].

did not perform an a priori search for such a smooth

transition in the phase measurements, no confidence

level can be derived from this result.

The infill surface detector array which is now op-

erating at the Pierre Auger Observatory will allow us

to extend this search for large scale anisotropies to

lower energy thresholds.

7 Serendipity observations and inter-

disciplinary science

The hybrid character of the Auger Observatory,

but also the surface and fluorescence detectors them-

selves allow a number of studies beyond cosmic ray

physics.

A first study, using the so-called scaler mode data

of the SD has been presented in [28] and is updated

at this conference [13, H. Asorey]. It uses the count

rates of low energy secondary cosmic ray particles (de-

posited energy ≥ 15 MeV and 15≤Edep ≤ 150 MeV,using two different triggers) recorded continuously for

self-calibration purposes for all of the 1660 SD sta-

tions. With each detector recording about 3600 Hz,

we thus record a total of about 6 MHz so that even

very small changes of the rates due to atmospheric

and solar changes can be monitored. This enables the

SD of Auger to address questions of solar cosmic rays

and allows to study Forbush events. A good agree-

ment between neutron monitor and the scaler data is

found when accounting for different geomagnetic cut-

offs of detectors located at different latitudes and for

different effective energy thresholds of neutron moni-

tors and the Auger SD stations [28].

Instead of using averaged scaler rates for the whole

array, it is also possible to study the scaler rate of

individual stations, in order to study the propaga-

tion of some phenomena across the Auger SD, like

the crossing of a storm over the 3000 km2 of the ar-

ray. This is because the flux of secondary particles

changes as the pressure front moves from across the

detector field. Additional analyses to study the influ-

ence of the variation of electric fields on the flux of

EAS particles are currently being carried out as well.

Interestingly, the 8.8 magnitude earthquake in

Chile on 27 Feb 2010 06h34 UTC with the epicen-

tre located about 300 km SW from the Auger Ob-

servatory left traces in the SD as well. The averaged

scaler rate for the whole array and also for individ-

ual stations showed a 24σ decrease beginning (90±2)seconds after the earthquake. This delay is compati-

ble with the propagation of seismic S-waves over that

distance. The scaler rate from 6h15 to 6h45 UTC is

Vol. 12, 64


shown in Figure 12. Although other minor quakes

have been recorded by seismographs near the SD, no

other similar effects have been found in 6 years of

data. Detailed analyses to identify the causes of the

observed drop in the scaler rate are underway. These

include simulations and shaking tests of selected de-

tectors in the array. After 6 hours, the scaler rate

recovered to the mean value for February 2010.

Also quite unexpectedly, during a normal FD data

taking shift an unusual event has been observed with

a well defined space-time structure: a luminous ring

starting from a cluster of pixels, and expanding in

all directions [29, A. Tonachini]. Usually, such kind

of events lasting for much longer than 70μs and with

such a high multiplicity are rejected by the T2 trig-

ger because of being caused by lightning with high

probability. Due to this rejection, only three of such

unusual events were recorded. By careful reconstruc-

tion of the timing, these events could be identified as

elves originating from lightning in the western part

of Argentinia. Elves are transient luminous phenom-

ena originating in the D-layer of the ionosphere, high

above thunderstorm clouds, at an altitude of approx-

imately 90 km. With a time resolution of 100 ns and

a space resolution of about 1 degree, the FD can pro-

vide an accurate 3D measurement of elves for thun-

derstorms which are below the horizon. To improve

the detection efficiency for such kind of interesting

and not well understood phenomena, a dedicated trig-

ger will be implemented in the future.

Fig. 12. Ten seconds average of the Auger

scaler rate for the 27 Feb 2010 Chile major

8.8 magnitude earthquake. A strong 24σ de-

crease is found 90±2 (stat) seconds afterwards,compatible with the time delay expected for

seismic S-waves traversing the distance from

the epicentre to the Auger Observatory.

Fig. 13. FD camera image for 4 consecutive time windows as indicated. It shows the time evolution of an

elve located at about 80 km altitude at a distance of 580 km from the observatory [29, A. Tonachini].

Vol. 12, 65


The Auger Observatory allows us to perform

a number of further interdisciplinary science stud-

ies, mostly related to atmospheric sciences (study of

aerosols, atmospheric gravity waves, etc.) but in-

cludes also biological studies in the pampa as well

as related studies of earthquakes either directly by

its instrumentation operated or indirectly by pro-

viding infrastructure for non-cosmic ray scientific

communities.

8 Summary and Conclusions

The Pierre Auger Observatory has reached a cum-

ulative exposure of more than 25 000 km2 sr yr by the

time of writing this article. This exceeds by far the

total statistics recorded by all other observatories. A

great deal of new insights are provided by these data,

but many new questions have appeared. This is most

prominently about the origin of the suppression of

the CR flux at highest energies and - related to this

- the mass composition and anisotropies at the high-

est energies. The Observatory will continue to collect

data with unprecedented precision for several more

years and it is hoped that these data will help to un-

ravel the puzzles about the most energetic particles

in nature.

9 Acknowledges

KHK acknowledges financial support by the Ger-

man Ministry of Research and Education, by the

Helmholtz Alliance for Astroparticle Physics (HAP)

and by DAAD.

References

[1] P. Abreu, et al., 2011, arXiv:1107.4809



[4] L. Cazon, R. A. Vazquez and E. Zas, 2005, Astropart.

Phys., 23, 393

[5] K.-H. Kampert and M. Unger, 2012, Astropart.

Phys., 35, 660

[6] R. Ulrich, et al., 2011, Phys. Rev. D, 83, 054026

[7] G. Aad, et al., 2012, Nature Comm., 2, 463

[8] R. Glauber, 1955, Phys. Rev., 100, 242

[9] R. Glauber and G. Matthiae, 1970, Nucl. Phys. B,

21, 135

[10] P. Bhattacharjee and G. Sigl, 2000, Phys. Rep., 327,

109

[11] G. Roos, et al., 2011, Astropart. Phys., 35, 140

[12] D. Heck, et al., 1998, “CORSIKA: A Monte Carlo

Code to Simulate Extensive Air Showers”, Report

FZKA, 1998, 6019.

[13] P. Abreu, et al., 2011, arXiv:1107.4805.

[14] G. Gelmini, O. Kalashev and D. Semikoz, 2008, J.

Exp. Theor. Phys., 106, 1061

[15] P. Abreu, et al., 2011, Phys. Rev. D, 84, 12205

[16] A. Cuoco and S. Hannestad, 2008, Phys. Rev.

D, 78, 023007; M. Kachelriess, et al., 2009, New

J. Phys., 11, 065017; L. A. Anchordoqui, 2011,

arXiv:1104.0509

[17] J. Abraham, et al., 2007, Science, 318, 938; J. Abra-

ham, et al., 2008, Astropart. Phys., 29, 188; P.

Abreu, et al., 2010, Astropart. Phys., 34, 314

[18] M.-P. Veron-Cetty and P. Veron, 2006, A&A, 445,

773

[19] T. Abu-Zayyad, et al., 2012, arXiv:1205.5984

[20] P. Abreu, et al., 2012, Astropart. Phys., 35, 354

[21] J. Abraham, et al., 2007, Astropart. Phys., 27, 244

[22] A. A. Abdo, et al., 2010, ApJS, 188, 405

[23] P. Abreu, et al., 2011, Astropart. Phys., 34, 627

[24] A. H. Compton and I. A. Getting, 1935, Phys. Rev.,

47, 817

[25] J. Candia, S.Mollerach and E. Roulet, 2003, JCAP,

0305, 003

[26] A. Calvez, A. Kusenko and S. Nagataki, 2010, Phys.

Rev. Lett., 105, 091101

[27] M. Kachelriess and P. Serpico, 2006, Phys. Lett. B,

640, 225

[28] P. Abreu, et al., 2011, J. Instr., 6, 1003


Vol. 12, 66

Documents

Highlights from the Pierre Auger Observatory · 2012. 12. 20. · Key words: Pierre Auger Observatory, UHECR, cosmic rays 1 Introduction and Status of the Ob-servatory The Pierre